BOEHLERT ET AL.; TIME SERIES OF GROWTH IN SEBASTES 



the prior upwelling season. The importance of 

 upwelling in rockfish feeding is shown by the 

 decreased fat storage in El Nino years in S. 

 flavidus (Lenarz and Echeverria 1986). 



For S. diploproa, wind stress curl was nega- 

 tively correlated with the gi-owth indexes for age 

 gi'oups 2, 4, and 6, with mean growth for age 

 gi-oups 2-6 (G26), and with the first principal 

 component (Table 1), which reflects most closely 

 gi-owth from ages 2-6 (Fig. 5B). Negative wind 

 stress curl is associated with offshore conver- 

 gence (Nelson 1977), which could concentrate 

 prey (such as euphausids) and potentially in- 

 crease feeding efficiency during the spring- 

 summer upwelling season when maximum feed- 

 ing takes place. Sea-surface temperature is 

 negatively correlated with growth in year 2, 

 whereas it is positively (but nonsignificantly) 

 correlated with growth in year 1; the correlation 



Table 1. — Pearson correlation matrices tor six age group 

 growth indexes (G1-6). mean growtti indexes (G for Se- 

 bastes pinniger. 626 for S. diploproa). and principal com- 

 ponent time series of ttie tfiree dominant modes of variability 

 (PC 1 -3) witti environmental factors from 1 946 to 1977: mean 

 yearly values (as opposed to smoottied values) are used in 

 ttnese calculations. The mean growth index for S. diploproa is 

 for ages 2-6 because of differences in growth in age group 1 . 

 The significance levels of coefficients were calculated follow- 

 ing Chelton (1983). CURL, wind stress curl at lat. 45°N, 

 long. 125 W: UP, yearly average Bakun upwelling index at 

 lat, 45 N; SSTNB, sea-surface temperature at Neah Bay, WA; 

 UP49, upwelling index averaged for April-September; and 

 SLSF, sea level at San Francisco. 



between the second principal component (which 

 reflects gi-owth in year 1) and sea-surface tem- 

 perature is significant, however, suggesting a 

 real difference between the response of these 

 age gi'oups to temperature. This difference may 

 be involved with an ontogenetic shift in the tem- 

 perature for optimum growth for this species as 

 noted by Boehlert (1981). 



With the exception of the correlations noted 

 above, the general lack of strong relationships of 

 fish growth with physical parameters may be 

 due to several causes. First, errors in methodol- 

 ogy may cause lower resolution in the growth 

 time series than would be necessary to demon- 

 strate the associations. Second, the complexity 

 of factors having an impact upon fish gi-owth 

 (Weatherley 1976; Brett 1979) makes detecting 

 relationships of growth with individual physical 

 factors difficult; furthermore, fish growth may 

 be related to environinental factors in a non- 

 linear fashion. Higher resolution in the measur- 

 ing technique and minimization of ageing errors 

 will be necessary to address the first of these 

 problems. The second problem can be addressed 

 through nonlinear analysis of the relationships 

 and continuation of the time series of gi'owth 

 anomalies, which will help define the relation- 

 ships of growth with physical and biological 

 factors. 



CONCLUSION 



The research we describe develops a new tech- 

 nique for constructing historical time series of 

 fish gi'owth and conducts analysis into the causes 

 of gi'owth variation in Sebastes diplopi-oa and S. 

 pinniger. The technique is apphcable to a variety 

 of marine organisms for which chronographic 

 records are deposited in some calcareous struc- 

 tures. Our analysis of the record for S. diploproa 

 and S. pinniger is characterized by a strong sig- 

 nal of increased growth late in the record. 

 Although this signal is confounded by questions 

 associated with size-selective mortality and 

 Lee's phenomenon, we interpret it to indicate a 

 density-dependent response to stocks depleted 

 by overfishing. Our interpretation can be tested; 

 if stocks of Sebastes are allowed to rebuild, the 

 dominant signal of increased growth evident in 

 the time series (Figs. 3, 4, 6, 7) should return to 

 more normal values. We recommend that these 

 time series be continued and that the techniques 

 be applied to other species. The interrelation- 

 ships among species and further investigation of 

 the role of environmental factors in annual fish 



803 



